Using stochastic simulations to connect models of crop evapotranspiration to observations from wireless sensors
Matthew R. Lurtz and Ryan R. Morrison
We describe the implementation of a cost-effective ecosystem-monitoring network and subsequently use the observations to inform well-known evapotranspiration equations to test if a model can replicate processes that generate the data. In this study, we designed a monitoring station that measures within-canopy temperature, soil-moisture presence at varying depths, and green-leaf temperature using a contactless infrared thermometer. We also incorporate non-water-limited and limited irrigation management schedules along with high-frequency nanosatellite imagery to bolster our data driven approach. The field analysis took place at an agricultural monitoring site in plots of corn that are fully irrigated and drought-stressed in eastern Fort Collins, CO. We build off the well-established theory that evapotranspiration can be linearly related to the difference between air temperature and leaf temperature. We hypothesize that our results comparing evapotranspiration in drought-stressed and fully irrigated conditions will have statistically significant differences in daily-integrated values of water use. We anticipate that by converting a deterministic evapotranspiration equation into a stochastic one, we will be able to derive a distribution of sensible heat flux and subsequently compare the simulated values with our in-situ observations. Preliminary results suggest that simplified estimations of evapotranspiration using a DIY sensor network are an acceptable alternative to more sophisticated methods of measuring evapotranspiration. The sensor network used in this study is a feasible alternative for measuring ecosystem water use and future research will test the applicability of the autonomous network in remote, natural ecosystems.